The clinical course of myasthenia gravis (MG) is variable, and spontaneous remission is still uncommon. Knowledge of the prognostic factors may help understand the course of MG and thus optimize its management. A systematic review search was conducted in MEDLINE and EMBASE for English language studies from 1985 through 2009. We identified additional studies by reviewing bibliographies of retrieved articles and hand search main journal of neurology. Studies evaluating variables associated with or predictive of remission in adult patients with MG were included. Because of methodological heterogeneity, we refrained from statistical pooling, instead, a best evidence synthesis was used for summarizing the results. From 1810 potentially relevant studies, 13 cohort studies met the inclusion criteria. The included studies were heterogeneous considerably in sample size, disease duration, follow-up years, definition of remission, and analysis. Study quality was limited by retrospective design in most studies and lack of multivariate analysis. Time of diagnosis from onset (<1 year) showed strong evidence of predicting a better remission. In studies using completely stable remission outcomes, there was strong evidence that age at onset (<40 years) was of prognostic importance. Furthermore, gender showed no association with remission. Time of diagnosis from onset and age at onset were found to be predictors of remission. Gender does not seem to predict the course of MG. Our findings should be interpreted with caution because of the clinical and methodological heterogeneity of included studies.